dsh-biocore 0.1.4 → 0.1.5

package/lib/index.js

@@ -167,6 +167,7 @@ export function readFullConfig(dataDir) {
     embedding: storedEmbedding(source),
     rerank: normalizeRerank(source.rerank),
     expression: normalizeExpression(source.expression),
+    sanitize_reply: source.sanitize_reply !== false,
   }
 }
 
@@ -176,6 +177,7 @@ function writeFullConfig(file, config) {
     embedding: config.embedding,
     rerank: config.rerank,
     expression: config.expression,
+    sanitize_reply: config.sanitize_reply !== false,
   }, null, 2) + '\n', { encoding: 'utf8', mode: 0o600 })
   try { chmodSync(file, 0o600) } catch { /* Windows may ignore POSIX modes */ }
 }
@@ -191,6 +193,7 @@ export async function handleSettings(req, res, dataDir) {
       embedding: maskedEmbedding(config.embedding),
       rerank: maskedRerank(config.rerank),
       expression: config.expression,
+      sanitize_reply: config.sanitize_reply !== false,
       restart_required: false,
     })
   }
@@ -213,7 +216,8 @@ export async function handleSettings(req, res, dataDir) {
   const hasEmbedding = body.embedding && typeof body.embedding === 'object'
   const hasRerank = body.rerank && typeof body.rerank === 'object'
   const hasExpression = body.expression && typeof body.expression === 'object'
-  if (!hasEmbedding && !hasRerank && !hasExpression) {
+  const hasSanitize = typeof body.sanitize_reply === 'boolean'
+  if (!hasEmbedding && !hasRerank && !hasExpression && !hasSanitize) {
     return sendJson(res, 400, { ok: false, reason: 'no_settings' })
   }
   if (hasEmbedding) {
@@ -236,12 +240,14 @@ export async function handleSettings(req, res, dataDir) {
     config.rerank = rerank
   }
   if (hasExpression) config.expression = normalizeExpression(body.expression)
+  if (hasSanitize) config.sanitize_reply = body.sanitize_reply === true
   writeFullConfig(paths.file, config)
   return sendJson(res, 200, {
     ok: true,
     embedding: maskedEmbedding(config.embedding),
     rerank: maskedRerank(config.rerank),
     expression: config.expression,
+    sanitize_reply: config.sanitize_reply !== false,
     restart_required: true,
     hint: '重启 BioCore 生效',
   })

package/lib/modules/core-turn.js

@@ -26,6 +26,11 @@ import { applyGateText, assistantTextOf } from './safety-gate.js'
 import { defaultStoreDir } from './trace.js'
 import { runtimeAction, runTurn } from '../py-bridge.js'
 import { EXPRESSION_DEFAULTS, readExpressionSettings } from './expression-config.js'
+import {
+  readSanitizeReplySetting,
+  sanitizeReply,
+  sanitizeReplyEnabled,
+} from './sanitize-reply.js'
 
 export const TURN_CONTEXT_NAME = 'biocore-expression-turn'
 export const TURN_CONTEXT_ORDER = -190
@@ -118,6 +123,7 @@ export function turnEndPayload(options) {
     store: opts.store || defaultStoreDir(),
     capture: false,
     memoryEnabled: false,
+    sanitize_reply: opts.sanitizeReply !== false,
     diagnostics,
   }, diagnostics, {
     modules_trace: diagnostics.modules,
@@ -135,6 +141,9 @@ export function makeCoreSessionListener(options) {
   const getExpression = typeof opts.getExpression === 'function'
     ? opts.getExpression
     : () => readExpressionSettings(stateDir)
+  const getSanitizeReply = typeof opts.getSanitizeReply === 'function'
+    ? opts.getSanitizeReply
+    : () => readSanitizeReplySetting(stateDir)
   const store = opts.store || defaultStoreDir()
   const log = typeof opts.log === 'function' ? opts.log : (line) => {
     try { console.error(line) } catch { /* noop */ }
@@ -186,9 +195,11 @@ export function makeCoreSessionListener(options) {
         }).catch(() => null)
         return
       }
-      // assistant/message: gate + trace via one turn_end call.
+      // assistant/message: sanitize -> gate + trace via one turn_end call.
       const original = assistantTextOf(event)
       if (!original) return
+      const sanitizeEnabled = sanitizeReplyEnabled(getSanitizeReply())
+      const cleaned = sanitizeEnabled ? sanitizeReply(original) : original
       const userText = lastUser.get(sessionId) || ''
       void (async () => {
         try {
@@ -198,11 +209,12 @@ export function makeCoreSessionListener(options) {
           const state = read && !read.error && !read.missing ? read.state : {}
           const payload = turnEndPayload({
             sessionId,
-            assistantText: original,
+            assistantText: cleaned,
             userText,
             switches,
             store,
             state,
+            sanitizeReply: sanitizeEnabled,
           })
           const result = await runAction('turn_end', payload)
           if (!result || result.ok !== true) return
@@ -216,6 +228,9 @@ export function makeCoreSessionListener(options) {
               + ' blocked=' + (gate.blocked === true ? '1' : '0')
               + ' rule=' + stringId(gate.rule || '')
               + ' chars=' + original.length + '->' + gate.text.length)
+          } else if (cleaned !== original) {
+            // No gate rewrite: still write the sanitized text back.
+            applyGateText(event, cleaned)
           }
           if (result.trace_file) {
             logLine(log, CORE_TRACE_PREFIX + ' session=' + sessionId

package/lib/modules/sanitize-reply.js

@@ -0,0 +1,60 @@
+/** Strip BioCore output markers from assistant replies (task-223).
+ *
+ * Removes:
+ *   - <bc>...</bc> and <biocore>...</biocore> blocks (also unclosed to EOL/EOS)
+ *   - bare compact JSON objects that only carry e/i[/n] keys
+ *   - extra whitespace; empty result stays empty (never falls back to input)
+ */
+import { readFileSync } from 'node:fs'
+import { join } from 'node:path'
+
+export const DEFAULT_SANITIZE_REPLY = true
+
+const TAG_RE = /<(bc|biocore)\b[^>]*>/gi
+const BARE_JSON_RE = /\{\s*"e"\s*:\s*"[^"\\]*(?:\\.[^"\\]*)*"\s*,\s*"i"\s*:\s*-?\d+(?:\.\d+)?(?:\s*,\s*"n"\s*:\s*-?\d+(?:\.\d+)?)?\s*\}/g
+
+function stripClosedTags(text) {
+  return String(text)
+    .replace(/<bc\b[^>]*>[\s\S]*?<\/bc\s*>/gi, '')
+    .replace(/<biocore\b[^>]*>[\s\S]*?<\/biocore\s*>/gi, '')
+}
+
+function stripUnclosedTags(text) {
+  // Until line end or a Chinese/ASCII sentence terminator; the terminator stays.
+  return String(text)
+    .replace(/<bc\b[^>]*>[\s\S]*?(?:[\n。！？!?]|$)/gi, '')
+    .replace(/<biocore\b[^>]*>[\s\S]*?(?:[\n。！？!?]|$)/gi, '')
+}
+
+function normalizeWhitespace(text) {
+  return String(text)
+    .replace(/[ \t]{2,}/g, ' ')
+    .replace(/[ \t]*\n[ \t]*/g, '\n')
+    .replace(/\n{3,}/g, '\n\n')
+    .trim()
+}
+
+/** Main entry: returns the cleaned reply (may be ''); never reverts to input. */
+export function sanitizeReply(text, options) {
+  const opts = options || {}
+  if (opts.enabled === false) return String(text == null ? '' : text)
+  const source = String(text == null ? '' : text)
+  let out = stripClosedTags(source)
+  out = stripUnclosedTags(out)
+  out = out.replace(BARE_JSON_RE, '')
+  return normalizeWhitespace(out)
+}
+
+/** Config.json top-level flag: only an explicit false disables cleaning. */
+export function readSanitizeReplySetting(dataDir) {
+  try {
+    const doc = JSON.parse(readFileSync(join(String(dataDir || '').trim(), 'config.json'), 'utf8'))
+    if (doc && typeof doc === 'object' && doc.sanitize_reply === false) return false
+  } catch { /* missing/bad config -> default on */ }
+  return DEFAULT_SANITIZE_REPLY
+}
+
+/** Payload helper: sanitize_reply defaults to true unless explicitly false. */
+export function sanitizeReplyEnabled(value) {
+  return value !== false
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-biocore",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "type": "module",
   "description": "BioCore DSH UI: per-session state tab, standalone instrument panel, /api/biocore/state, session/event one_turn updates, /bio sync.",
   "main": "lib/index.js",

package/ui/panel.html

@@ -247,6 +247,7 @@
       <dl class="advanced-info">
         <div><dt>数据目录</dt><dd id="dataDirText">宿主托管</dd></div>
         <div><dt>插件版本</dt><dd id="pluginVersion">0.1.0</dd></div>
+        <div><dt>出口清洗</dt><dd id="sanitizeReplyText">开</dd></div>
       </dl>
       <div class="advanced-actions">
         <button id="resetLayout" class="advanced-button" type="button">恢复默认布局</button>

package/ui/panel.js

@@ -1050,6 +1050,7 @@
           fillModel(embedding)
           fillRerank(body.rerank)
           fillExpression(body.expression)
+          setText('sanitizeReplyText', body.sanitize_reply === false ? '关' : '开')
           setModelStatus('已载入')
           return
         }